Large language models have greatly enhanced our ability to understand biology and chemistry, yet robust methods for structure-based drug discovery, quantum chemistry and structural biology are still sparse. Precise biomolecule–ligand interaction datasets are urgently needed for large language models. To address this, we present MISATO, a dataset that combines quantum mechanical properties of small molecules and associated molecular dynamics simulations of ~20,000 experimental protein–ligand complexes with extensive validation of experimental data. Starting from the existing experimental structures, semi-empirical quantum mechanics was used to systematically refine these structures. A large collection of molecular dynamics traces of protein–ligand complexes in explicit water is included, accumulating over 170 μs. We give examples of machine learning (ML) baseline models proving an improvement of accuracy by employing our data. An easy entry point for ML experts is provided to enable the next generation of drug discovery artificial intelligence models.

The parasitic protists of the Trypanosoma genus infect humans and domestic mammals, causing severe mortality and huge economic losses. The most threatening trypanosomiasis is Chagas disease, affecting up to 12 million people in the Americas. We report a way to selectively kill Trypanosoma by blocking glycosomal/peroxisomal import that depends on the PEX14-PEX5 protein-protein interaction. We developed small molecules that efficiently disrupt the PEX14-PEX5 interaction. This results in mislocalization of glycosomal enzymes, causing metabolic catastrophe, and it kills the parasite. High-resolution x-ray structures and nuclear magnetic resonance data enabled the efficient design of inhibitors with trypanocidal activities comparable to approved medications. These results identify PEX14 as an “Achilles’ heel” of the Trypanosoma suitable for the development of new therapies against trypanosomiases and provide the structural basis for their development.

The COVID-19 pandemic caused by SARS-CoV-2 has been socially and economically devastating. Despite an unprecedented research effort and available vaccines, effective therapeutics are still missing to limit severe disease and mortality. Using high-throughput screening, we identify acriflavine (ACF) as a potent papain-like protease (PL^pro) inhibitor. NMR titrations and a co-crystal structure confirm that acriflavine blocks the PL^pro catalytic pocket in an unexpected binding mode. We show that the drug inhibits viral replication at nanomolar concentration in cellular models, in vivo in mice and ex vivo in human airway epithelia, with broad range activity against SARS-CoV-2 and other betacoronaviruses. Considering that acriflavine is an inexpensive drug approved in some countries, it may be immediately tested in clinical trials and play an important role during the current pandemic and future outbreaks.

The presence of an unpaired electron in paramagnetic molecules generates significant effects in NMR spectra, which can be exploited to provide restraints complementary to those used in standard structure-calculation protocols. NMR already occupies a central position in drug discovery for its use in fragment screening, structural biology and validation of ligand–target interactions. Paramagnetic restraints provide unique opportunities, for example, for more sensitive screening to identify weaker-binding fragments. A key application of paramagnetic NMR in drug discovery, however, is to provide new structural restraints in cases where crystallography proves intractable. This is particularly important at early stages in drug-discovery programs where crystal structures of weakly-binding fragments are difficult to obtain and crystallization artefacts are probable, but structural information about ligand poses is crucial to guide medicinal chemistry. Numerous applications show the value of paramagnetic restraints to filter computational docking poses and to generate interaction models. Paramagnetic relaxation enhancements (PREs) generate a distance-dependent effect, while pseudo-contact shift (PCS) restraints provide both distance and angular information. Here, we review strategies for introducing paramagnetic centers and discuss examples that illustrate the utility of paramagnetic restraints in drug discovery. Combined with standard approaches, such as chemical shift perturbation and NOE-derived distance information, paramagnetic NMR promises a valuable source of information for many challenging drug-discovery programs.